sign in sign up
<<
Home , Hafting, Howiesons Poort, Sibudu Cave

Journal of Evolution 49 (2005) 587e601

Putting to the test: replication studies of adhesives that may have been used for in the Middle

Lyn Wadley

Archeology Department, University of the Witwatersrand, PO WITS 2050,

Received 14 April 2005; accepted 22 June 2005

Abstract

Substantial frequencies of (MSA) lithics from Rose Cottage and Sibudu in South Africa have red ochre on their proximal and medial portions. Residue studies suggest that the tools were hafted and that the ochre may be part of the adhesive used for hafting the tools. Replication studies show that ochre is indeed a useful loading agent for adhesive; however, there are other potential loading agents. It is also possible to use unloaded plant resin, but this agent is brittle and difficult to work with. It appears that people living in the MSA had wide knowledge of ingredients suitable for hafting tools, and that they chose different adhesive recipes because of the required properties of the adhesive. Brittle, unloaded adhesive allows a projectile head to disengage its haft and implant itself in an animal; robust adhesive keeps a spearhead safely in its shaft. Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: South Africa; Middle Stone Age tools; Hafting; Red ochre; Adhesive; Replication

Introduction The positions of the ochre suggest at least three reasons for its presence: (1) some pieces are ochre- This study has its roots in the observation, made covered because they fortuitously rested on ochre- some years ago, that many stone tools from Rose stained ground, (2) others seem to have been used to Cottage , South Africa, have red ochre on cut or scrape ochre, and (3) numerous tools have them. Residue and macrowear studies of these ochre on their bases. It is this latter group that I MSA tools and similar tools from , concentrate on in this paper. My working hypoth- also in South Africa, are described in this paper. esis is that many tools were hafted using an adhesive in which red ochre was an ingredient. My hafting E-mail address: [email protected] replication studies were thus designed to investigate

0047-2484/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jhevol.2005.06.007 588 L. Wadley / Journal of 49 (2005) 587e601 the usefulness of ochre as an ingredient in natural symbolic behavior (e.g., Knight et al., 1995; adhesives. If the ochre can be shown to have little Deacon, 1995; McBrearty and Brooks, 2000; Watts, purpose, then it is feasible that it was added to the 2002), and others (e.g., Deacon, 1995; Wurz, 1999) adhesive for the sake of its color and possibly for consider backed tools to be hallmarks of modern symbolic purposes. Naturally, a symbolic role behavior, embodying symbolic expression. If the cannot be totally excluded if the ochre is found to ca. 300 kyr-old ochre and backed tools are indeed be an advantageous additive to adhesives, but this attributes of a symbolic system, then symbolism is scenario will suggest that it is unwise to automat- not the exclusive property of anatomically modern ically assume symbolism when ochre is present. , and archaeologists will need to redefine Coloring material is present early in African what it means to be a culturally modern human. living sites: it was used systematically by about The common assumption about the use of large 285 ka at GnJh-15, Kapthurin Formation, quantities of ochre in the MSA is that red coloring (McBrearty, 2001), and possibly even earlier at was used for symbolic body decoration. This Twin Rivers, , where were found conjecture is based on analogy with modern associated with a Lupemban , dating to hunter-gatherers living tens of thousands of years about 300 ka (Barham, 2002). Barham (2002: 186e later. For example, red body-paint was used 187) estimated that about 60 kg of coloring material for puberty and marriage rituals by !Kung girls were recovered from the 1950s excavations. This in northwestern in historic times large quantity of coloring is overshadowed by the (Marshall, 1976: 277). There is, however, no way tons of specular hematite that were mined from iron of testing whether ancient people practiced body pods at Lion Cavern, in the eastern part of South painting, and the application of hunter-gatherer Africa, perhaps earlier than 40,000 years ago. ethnography to the deep past can be problematic; Mining hammers in the form of grooved, heavy- as pointed out by Hovers et al. (2003), ethno- duty stones were found alongside Middle Stone Age graphic analogy cannot be used as both a building (MSA) tools in one of the Lion Cavern adits block of a model and a test of the same model. (Beaumont, 1973: 140). It is difficult to imagine Powdered ochre was sometimes placed on the what people were doing with this amount of bodies of the dead in southern African Later Stone coloring material, but it looks as if the demand for Age (LSA) sites dating to within the last 10,000 coloring material increased between the Lupemban years (Wadley, 1997), but symbolic behavior is and the final MSA. I address the issue of densities of already -documented by this time. In my coloring material at the end of this paper. opinion, the earliest unequivocal evidence for Not only does the Lupemban Industry document symbolic behavior in Africa comes from Blombos the earliest substantial use of ochre in Africa, but it Cave, in the Western Cape, where shell were also has the earliest backed tools. At Twin Rivers, found in a layer that is dated to approximately backing accounts for up to 15% of the retouched 77 ka (Henshilwood et al., 2004). These beads are tools in the Lupemban Industry (Clark and Brown, symbolic because they had the potential to play an 2001; Barham, 2002). It might not be coincidental active role in the articulation and manipulation of that backed tools and quantities of ochre are found social relationships (Wadley, 2001, 2003). As together; the reason for this comment will be made personal ornaments, they were an index of social clear below. The Lupemban industry appears to be identitydeither group or individual identity. Such associated with fossil remains of Homo heidelber- expressions of identity are always symbolic. The gensis (elsewhere called Homo rhodesiensis)at earliest use of ochre for no clear functional reason Kabwe, Zambia (Barham, 2002). There are impor- comes from the same 77 kyr-old layer in Blombos tant implications if these anatomically pre-modern Cave (Henshilwood et al., 2002). Here, several humans were the makers of the backed tools and the pieces of engraved ochre were found. The users of ochre at Kabwe and at Twin Rivers (and larger Blombos ochre tablet has a crosshatched possibly Kalambo Falls). Some archaeologists design engraved inside several broken boundary believe that ochre is always associated with lines. The decorated ochre was interpreted by L. Wadley / Journal of Human Evolution 49 (2005) 587e601 589

Henshilwood et al. (2002) as evidence for the reason for the use of red rather than other colors of presence of symbolic behavior and, therefore, ferruginous material. The extinct /Xam from the cultural modernity at 77 ka. The engraved tablets northwestern part of South Africa used red ochre to are certainly compelling evidence for new types of color leather bags (Bleek and Lloyd, 1911); the behavior in the MSA. Since the shell beads are custom of coloring some leather bags red continues already evidence for symbolism at 77 ka, we know in parts of Botswana and (personal that the engraved ochre was made by people who observation). It is also correct to say that there are used symbolism. This means that the ochre may several methods for successfully tanning leather and have had a symbolic function, but, in my view, the that ochre is not necessarily a requirement. South- meaning of the engraved ochre remains ambigu- ern Ethiopian workers, for example, use plain ous. More than 8000 unworked pieces of ochre water on their skins (Kimura et al., 2001). Nonethe- were recovered from the same site, so ochre less, ochre is a valuable tanning agent. Steinman production was common at Blombos (Henshil- observed Tehuelches in tanning guanaco wood et al., 2002). If the ochre was not always for hides with ochre and fat in 1906, and Sollas symbolic purposes, what could the non-symbolic described ochre-assisted hide tanning in Tasmania functions have been? In particular, what could red (in Audouin and Plisson, 1982: 57). Audouin and ochre’s functions have been before 77 ka, i.e., at Plisson (1982) were able to demonstrate that hide 300 ka when quantities of ochre were being processing with red ochre is beneficial, particularly brought back to living sites like Twin Rivers? for preventing or reversing decomposition. They Henshilwood and Marean (2003: 635) stated treated a three-day-old moose skin when it was that Watts (2002) had effectively critiqued the already in the first stages of putrefaction. The skin suggestion that ochre was utilitarian. This opinion was scraped with flakes, and red ochre was then is shared by several archaeologists, notwithstand- applied everywhere, except the tail. The hide dried ing the evidence from several continents that ochre quickly and became thinner and softer, but the tail has properties that make it useful for practical section that was not covered in ochre rotted. Thus, tasks, such as hide-working and adhesive manu- red ochre seems to be especially desirable for facture. This claim does not detract from ochre’s tanning leather where there is a risk of decay. worth as a symbolic agent, and it does not deny Another of Audouin and Plisson’s experiments that, in the recent past, many tasks that on the involved tanning two pieces of skin from the same surface appear practical (such as hunting) can be ox hide. The piece tanned with yellow ochre imbued with symbolism (Wadley, 1987). This (goethite) stayed stiff, thick, and rough, whereas claim also does not mean that other substances the piece worked with red ochre dried rapidly and cannot be used equally well for hide-working or became pleasingly thin and soft. Audouin and adhesive manufacture. People in the past, like Plisson deduced that red ochre stops hide from people today, probably chose from a variety of rotting, assists it to dry rapidly, and produces materials when they were carrying out tasks. superior leather compared to yellow ochre. Goe- Little has been written about the functional uses thite is a hydrated iron oxide, and therefore has of ochre, but ochre has been shown to have higher moisture content than red ochre. properties that make it an antibacterial Replication work (Allain and Rigaud, 1986; agent (Mandl, 1961: 196; Velo, 1984). In desert Wadley et al., 2004) and microscopy (Williamson, , fruits of Solanum sp. and the wild fig were 1997, 2004; Gibson et al., 2004; Lombard, 2004, packed into balls of ochre the size of a basketball 2005; Wadley et al., 2004) have demonstrated that and were stored in trees to conserve them (Flood, ochre has at least one other function in addition to 1995: 264). This practice confirms a preservative its effectiveness as a preserving agent and a tanning role for ochre. Red ochre reduces collagenase and is agent. Replication studies in by Allain and therefore perfect for processing hides (Audouin and Rigaud (1986: 715) demonstrated that successful Plisson, 1982). Watts (2002) argued that ochre is not adhesives depend on a loading agent in the form of needed for tanning, and that there is no practical an inert powder, such as ochre. Wax and resin do 590 L. Wadley / Journal of Human Evolution 49 (2005) 587e601 not mix well without a loading agent, which also surface of the site contains debris from encourages the hardening of the adhesive while it occupations; immediately below this are multiple dries. The combined ingredients prevent adhesives MSA layers containing final MSA dating to about from becoming brittle. Allain and Rigaud’s recipe 40 ka (Wadley, in press), late MSA dating to uses one part beeswax, four parts resin and one between 60 ka and 40 ka (Villa et al., 2005), part yellow ochre, and the mixture is heated in (Wadley and Jacobs, 2004), order to melt the components. A well-known and pre-Howiesons Poort MSA industries. In property of goethite is that it readily dehydrates a preliminary analysis of 412 late MSA tools (no and transforms to hematite with a corresponding backed tools were represented here), Williamson color change when it is heated at temperatures as (2004) discovered that 29 out of 104 points from low as 250 C(Pomie` s et al., 1999). This means the late MSA layers had ochre residues on them, that archaeologists can expect to find red rather 30 out of 83 scrapers, 23 out of 77 pieces of than yellow adhesives when ochre has been used as ‘‘other’’ retouch, and 26 out of 113 flakes had a loading agent. Pomie` s et al. (1999) suggested that ochre on them. Although the percentages of ochre yellow ochre that is transformed through heat to in each class are similar, the position of ochre a darker color can be structurally distinguished on tools within the various classes is not the same. from its unheated hematite counterpart, even For example, only 3% of scrapers have ochre on when the color is identical. Such a study could their working edges, whereas 27% of flakes have be profitable on archaeologically recovered tools. ochre residues on their working edges (Wadley The French adhesive replication experiments et al., 2004). Some flakes may have been used for were initiated because some French tools processing ochre and, in some instances, for have ochre on them and this required explanation processing items such as leather together with (Audouin and Plisson, 1982; Beyries and Inizan, ochre. Some flakes appear to have been hafted 1982). A backed with ochre on its backed edge because 47% have ochre on their proximal and/or was recovered from the Gouy in medial portions. The proximal end of a tool northern France and, at in southwest incorporates the platform and bulb of percussion. France, adhesive containing red ochre was noticed In contrast to flakes, 80% of scrapers and 68% of on backed bladelets (Audouin and Plisson, 1982: points have ochre on their proximal and/or medial 52). Ochre has been found on stone tools from portions. The microscopic analysis of the Sibudu a further eight French sites, but the position of the tools with ochre residues shows that, on retouched ochre is not documented (Allain and Rigaud, 1986). tools of all classes, the majority of the ochre In a preliminary residue analysis of artifacts from residues occur on proximal, medial, or a combina- Rose Cottage Cave, South Africa, Williamson tion of these two positions. (1997) detected the presence of red ochre on many The tools for residue analysis were usually stone tools. A subsequent study of Howiesons Poort taken directly from their excavated position with Industry backed tools from the MSA of the same sterilized tweezers, and a small soil sample was site showed that ochre and plant material were often taken at the site where the tool was recovered. The concentrated together on or near the backed edges soil samples were analyzed so that it can be seen (Tomlinson, 2001; Gibson et al., 2004). This co- whether residues on the tools could have resulted incidence suggested that ochre might be part of the from soil contamination (Williamson, 2004). The hafting process, but the sample was small and was tools used for the residue analyses presented here not ideal for residue work because the Rose Cottage were not contaminated with ochre from the soil. tools had been handled and labeled after excavation However, patches of ground ochre were sometimes in the 1980s. It was therefore decided to study found in Sibudu MSA deposits, and fortuitously untouched MSA stone tools from Sibudu Cave. ochre-impregnated tools and bones were always Sibudu Cave is approximately 12 km inland recovered from these patches. Such tools were from the Indian Ocean and about 40 km north of never used for residue analysis. When conducting in KwaZulu-Natal, South Africa. The residue analyses, it is important to understand the L. Wadley / Journal of Human Evolution 49 (2005) 587e601 591 exact provenience and the context of the collection ochreous dust and sand with their resins (Rots, that is being studied. 2002: 60). San from Namaqualand used ‘‘eu- With only a few exceptions, there are more phorbid milk and red clay’’ as ‘‘cement’’ for plant residues on tools that have ochre on their attaching feathers to according to Doctor bases than on tools without ochre. This may imply Atherstone, who made the observation in 1854 that the tools were secured with plant twine in (Webley, 1994: 61e62). The red clay may have addition to adhesive. been ochre. These ethnographic records should not In a detailed analysis of 24 whole points with be used as supportive evidence for similar ochre ochre residues from the late MSA of Sibudu, use in the past, but the cross-continental use for Lombard (2004, 2005; Wadley et al., 2004) ochre in adhesives is, nonetheless, interesting. observed clusters of bending fractures on the Red ochre was sometimes replaced with other laterals within the ochre-stained proximal and/or ingredients that acted as loading agents. Later medial portions of the points. Such fracture Stone Age (LSA) adhesives are not recorded as damage is consistent with hafting, and it probably having ochre in them (Phillipson, 1976; Schweit- results, in part, from the use of binding materials. zer, 1979; Deacon, 1979; Deacon and Deacon, Unfortunately no tools from Sibudu have been 1980; Binneman, 1983; Binneman and Deacon, found with large lumps of adhesive on them, but we 1986; Mitchell, 1995; Jerardino, 2001). Jerardino believe that the multifaceted evidence from the (2001) noted that the adhesive used for the residue and macrowear analyses provide compel- mounting of a stone from Steenbokfontein ling circumstantial evidence for hafting. The Cave contained plant fiber (Jerardino, 2001: 862), combination of ochre and damage to the lower but no ochre was observed. half of the points, sometimes in combination with As a preliminary observation, it therefore seems resin and other plant residues, implies that these that ochre is a useful loading agent for adhesive tools were hafted and that ochre was one of the manufacture, but that it can be replaced with other ingredients of the adhesive (Wadley et al., 2004). ingredients. My replication studies, which I shall While I assume that the staining of red ochre on the now describe, were designed to investigate whether bases (as opposed to the working edges) of stone the addition of ochre to local resin is as tools is most often the result of incorporating ochre constructive as it was shown to be in France. into adhesives that were used to mount the tools onto handles, this may not always have been the case. Other methods of hafting with ochre are Methods possible: Bu¨ller (1988), for example, suggested that tools in the Near East were ochre-stained because Nodules of hematite (approximately three kilo- they were wrapped in ochre-stained hide and then grams of nodules) and coarse sandstone slabs were hafted to bone handles. Not all tools with evidence collected from a quarry in the Wildlife Sanctuary, for hafting in the form of edge damage have ochre Moletadikgwa, Limpopo Province, South Africa. traces on their bases; this suggests that adhesive Resin from Acacia karoo trees and branches from recipes without ochre also existed in the past. Grewia flava shrubs were collected from the same Ochre has been observed on some stone tools sanctuary. Fibrous leaves of the plant Hypoxis from other sites, such as the MSA site of Die rigidula were picked to use for binding material. Kelders, Western Cape, South Africa (Thackeray, Wooden hafts were prepared by cutting a split 2000: 157). In Kenya, backed blades from Enka- in the haft or by cutting an L-shaped platform of pune Ya Muto, dated to between 50,000 and about 12 mm in width and 18 mm in height. 40,000 years ago, have red ochre residues on their Replicated flakes were made from a variety of backed portions (Ambrose, 1998). Historical local rock types, and 38 flakes were hafted with records suggest that ochre was also recently used replicated adhesives. as a loading agent for adhesives in Australia, The hematite formed a centimeter thick crust, where aborigines combine vegetal fiber and as the result of weathering, around an inner core of 592 L. Wadley / Journal of Human Evolution 49 (2005) 587e601 ironstone. Powdered coloring material was pro- 3. The adhesive was used to mount the stone duced by rubbing each nodule on a sandstone slab. flake on the haft, the finished tool was heated The nodule is most easily worked by rotating it so in the fire, and was then air dried. that several faces are ground. This method 4. The adhesive was used to mount the stone produces ground, striated facets on the nodule, flake on the haft; the finished tool was heated which is worked to the inner core until no further in the fire, and was then dried close to a fire. coloring can be ground. Consequently, much of 5. The adhesive was heated on the stirring stick each nodule is discarded, even though it still looks over the fire. The adhesive was used to mount red. The three kilograms of nodules produced the stone flake on the haft, and the finished sufficient hematite powder (about 70 ml) to make tool was air dried. adhesive for 28 flakes. The grinding process results 6. The adhesive was heated on the stirring stick in widespread reddening of everything in a meter- over the fire. The adhesive was used to mount wide radius of the activity. Hands, arms, clothing, the stone flake on the haft, and the finished work surface, and anything lying nearby become tool was heated in the fire and then air dried. covered in fine, red dust. 7. The adhesive was heated on the stirring stick Each tool was attached to its haft with over the fire. The adhesive was used to mount approximately 5 ml of adhesive. Four basic the stone flake on the haft, and the finished adhesive recipes were used: tool was heated over the fire and then dried close to the fire (Fig. 1a). 1. Acacia karoo resin (5 ml per tool) 2. Acacia karoo resin (2.5 ml) mixed with equal Some of the air-dried tools had Hypoxis twine quantities of red coloring powder (2.5 ml) wrapped around the adhesive while it was still wet, 3. Acacia karoo resin mixed with equal quantities but most were unbound (Fig. 1b, c.). All of the of red coloring powder and beeswax completed tools were photographed, and were 4. Acacia karoo resin mixed with equal quantities subsequently used for the same task of chopping of red coloring powder and a few drops of bark from branches for six minutes. The surfaces water of the various adhesive recipes were microscopi- cally examined at magnifications ranging from Acacia resin was selected because it is widely 50! to 500! under incident light. used as adhesive in Africa (Grant and Thomas, The hafting experiments were carried out on 1998; Van Wyk and Gericke, 2000); in southern four separate occasions (in June, August, and Ethiopia, stone scrapers for hide-working are still December 2003, and January 2004), and the mounted in adhesive that contains hardwood hafting techniques varied according to the results resins (Kimura et al., 2001). Acacia spp. occur obtained and lessons learned from each set of near Sibudu Cave today and they also occurred in experiments. Methods used are listed in Table 1. the area in the MSA (Allott, 2004; Wadley, 2004), so their resin would have been available through- out MSA occupation of the site. Results The processing of the adhesive recipes varied. In all cases adhesive was mixed with a stirring stick. Both split hafts and L-shaped hafts were used in Thereafter, the methods differed. the first replication. Although it is counterintui- tive, the L-shaped hafts were far more successful, 1. The adhesive was used to mount the stone even without binding. The split hafts were un- flake on the haft, and the finished tool was air successful because pressure on them extended the dried. split and the tool then fell from the haft (only one 2. The adhesive was used to mount the on a split haft did not break). Consequently, flake on the haft, and the finished tool was all split-haft results were excluded from the dried close to a fire. calculations in this section. Extensive binding can L. Wadley / Journal of Human Evolution 49 (2005) 587e601 593

reason, binding the tools with Hypoxis twine was also discontinued by the fourth experiment. Apart from the issue of unwanted variability, the marks made by the twine on the adhesive interfered with microscopic analysis of the adhesive. Twine is, however, most likely to have been used to strengthen hafts. Lombard’s (2004, 2005) micros- copy showed wear traces on a sample of points from Sibudu Cave that are most likely to have resulted from binding with twine.

Adhesive recipes

1. Natural plant resins. The quality of A. karoo resin seems to be variable, but it is difficult to quantify this suggestion. Certainly, drier gums are easier to work with than runny ones, which are very sticky. Unloaded resin is difficult to control during the hafting process: when it is the consistency of honey it requires constant attention to prevent it from dripping off the haft. For the same reason, the stirring stick must be rotated constantly and quickly while the resin is being dried over the fire. Even after the moisture content of the resin has been sufficiently reduced to allow the resin to be molded around the stone and its haft, the tool must be turned every half minute while it is drying next to the fire. If this is not done, the adhesive oozes off the haft. Natural plant resin dries slowly and the end product is brittle, full of cracks and air bubbles, and sometimes crumbly. Under the microscope, natural resin adhesive has a glassy, brittle appear- ance (Fig. 2). Woody and other plant inclusions are also apparent in the resin; some of these inclusions are unintentional and are from the scraping of gum from the bark of Acacia trees. Thus, plant inclusions in the hafting area of Fig. 1. (a) Hafted tools drying near a fire; (b) Haft 40 (resin and ochre mastic); (c) Haft 24 (resin, ochre and wax mastic, bound archaeologically recovered stone tools need not with Hypoxis twine). imply the deliberate use of non-resinous plant material in the adhesive recipes. Natural resin is water-soluble, and hydroscopic sometimes prevent the split from developing, but and damp conditions will cause adhesive to split hafting was nevertheless abandoned after the become tacky, with the result that the lithic insert first replication. The purpose of the study was to will fall from its haft. Haft 7 is a good example. examine the strength of various adhesive recipes; When the tool broke from its haft, it was put into varying the haft design introduced unwanted a plastic bag together with its haft. Within two variability into the experiments. For the same days the resin had become damp and sticky. 594 L. Wadley / Journal of Human Evolution 49 (2005) 587e601

Table 1 Hafting replication methodology and results

Haft# Haft Resin Ochre Wax H2O Heat/ Heat/fire/ Tool Dried Bind Description Task Status type stove stirring heated next of time stick on fire to fire mastic 1 Split Yes Yes eeYes eeeecrumbly 1 min broke 2 L Yes Yes eeYes eeee 1 min broke 3 Split Yes Yes eeYes eeee 5 min OK 4 L Yes Yes e Yes Yes eeeYes 5 min OK 5 L Yes Yes e Yes Yes eeeedidn’t set e 6 Split Yes eeee e e eebrittle 1 min broke 7 L Yes eeee e e ee 5 min broke 11 Split Yes* Yes e Yes Yes eeeee esplit 12 Split Yes Yes eeYes eeeee esplit 13 Split Yes* Yes eeYes eeeecrumbly e broke 14 Split Yes eeee e e ee e e e 15 L Yes Yellow e Yes Yes eeeee!1 min broke 16 L Yes Yes eeYes eeeee6 min OK 17 L Yes Yes eeYes eeeee ee 18 Split Yes Yellow eeYes eeeecrumbly ee 19 Split Yes Yes ee e e e e e sticky e 21 L Yes Yes ee e Yes Yes e Yes crumbly 1.5 min broke 22 L Yes Yes Yes ee Yes Yes e Yes elastic 6 min OK 23 L Yes Yes Yes ee Yes Yes e Yes elastic 6 min loose 24 L Yes Yes Yes ee Yes Yes e Yes elastic 6 min OK 25 L Yes Yes ee e Yes eeYes hard 6 min OK 26 L Yes eeee Yes Yes Yes crumbly 6 min OK 27 L Yes Yes ee e Yes eeYes hard 6 min OK 28 L Yes eeee e Yes e Yes brittle 6 min crumbly 29 L Yes Yes e Yes e Yes Yes e Yes hard 6 min OK 30 L Yes Yes ee e Yes Yes e Yes hard 6 min OK 31 L Yes Yes e Yes e Yes Yes e Yes hard 6 min OK 32 L Yes Yes ee e Yes Yes e Yes hard 6 min OK 33 L Yes Yes ee e Yes Yes e Yes hard 6 min OK 34 L Yes# eeee e e eYes hard 6 min OK 40 L Yes Yes ee e e Yes Yes e hard 5 min OK 41 L Yes Yes ee e e Yes Yes e hard 5 min OK 42 L Yes eeee e Yes Yes e hard 5 min OK 43 L Yes Yes ee e Yes e Yes e hard 5 min OK 44 L Yes eeee Yes e Yes e hard 5 min OK 45 L Yes Yes ee e Yes e Yes e hard 5 min OK 46 L Yes Yes ee e Yes Yes Yes e hard 5 min OK 47 L Yes Yes ee e e e Yes e hard 5 min OK * Elephantorrhiza resin. # Ozoroa resin.

Notwithstanding the problems that I have out- successful (Haft 28) survived the tasks, but lost so lined, the replications show that A. karoo resin can much of its brittle adhesive that it cannot be used for be used alone as adhesive for attaching tools to their any further work. Other vegetable resins, e.g., hafts. Of the five tools mounted with Acacia gum, Ozoroa paniculosa, are also successful adhesives; four were successful for cutting tasks that lasted five Haft 34 demonstrates this. Elephantorrhiza burkei or six minutes (Table 1). This corresponds to an gum was very crumbly and was not successful. 80% success rate, but the sample may be too small 2. Plant resin mixed with red ochre. Fifteen of to be representative. One of the tools labeled the red ochre and Acacia gum adhesives on L-hafts L. Wadley / Journal of Human Evolution 49 (2005) 587e601 595

were successful, while three were not. This represents an 83% success rate. One broken tool (Haft 21) was air dried and had not set sufficiently before it was used; when it broke, the inner adhesive was still wet. I comment on correct drying methods later. The other two unsuccessful adhesives had water mixed with them during manufacture. As a result, the adhesive on Haft 5 did not set and it could not be used. This happened because, in my first set of experiments, I mixed and heated more adhesive than was needed for the hafting of a single tool. When molding of the first tool was finished, the remnant adhesive was crumbly, so a few drops of water were added to moisten the mixture. Although the reconstituted adhesive was reheated and initially appeared successful, it never dried. The addition of water to the adhesive was not, however, always un- satisfactory. When water was mixed with fresh Acacia gum and red coloring in the first stage of manufacture, it formed successful glue after it was heated. Indeed, this combination held tools so well during scraping and cutting of wood that the stone inserts could not be pried from their hafts (Hafts 29 and 31). Thus, all three of the unsuccessful resin/ochre adhesives appear to have failed be- cause of a recipe error. Because the natural plant resins vary in their moisture content, a set recipe cannot be prescribed for the addition of a loading agent. Wet resins require more loading agent than drier resins in order to make them manageable. Thus, the making of adhesive is more complex than might be expected, and experience is required to get an acceptable outcome. When red ochre/resin adhesive has been prop- erly dried, it does not seem to be adversely affected by damp conditions or by storage in plastic. In this respect, it also has a big advantage over pure resin adhesive. The adhesive is, however, water-soluble when soaked for several hours. Under the microscope, the adhesives loaded with red ochre appeared homogeneous (Fig. 2), although cracks sometimes developed; fine ‘‘mud cracking’’ (Fig. 1b) occurred as the result of slow drying next to the fire.

Fig. 2. (a) Ochre and resin mastic (Haft 30); (b) heated Acacia resin; (c) ochre, resin, and wax mastic (Haft 24). 596 L. Wadley / Journal of Human Evolution 49 (2005) 587e601

3. Plant resin mixed with red ochre and wax. fire for creating and ‘‘’’ the adhesive is as Only three tools were made using a combination of much an art as the mixing of ingredients. On my equal quantities of resin, red ochre, and natural first attempt at ‘‘cooking’’ a newly hafted tool over beeswax. All three tools were successful, although the fire, the adhesive swelled and the outer crust Haft 23 became loose after it was used for six became hard and charred, leaving an air-filled minutes. The addition of wax made the adhesive hollow under the hard crust. Even without this especially easy to work with; it had the consistency incompetent handling, heating of adhesive over and flexibility of plasticine. The waxy adhesive the fire can introduce charcoal into it. This may used in my experiments never set completely hard; account for the charcoal that is sometimes a fingernail could be pushed into the adhesive observed (B. Williamson, personal communica- several months after its manufacture. The wax- tion) on the bases of tools during microscopy. impregnated adhesive was also very homogeneous There is, however, another important application and had few cracks. The disadvantage of using for heat in the next stage of adhesive production. wax is that shrinkage occurs within a period of The slow drying of the completed tools near several weeks, and this caused the adhesive to a moderate fire makes the drying process consid- loosen from its haft. Shrinkage could possibly be erably quicker than drying with no additional avoided by using less wax. Allain and Rigaud heat. The difficulty experienced with air drying (1986) used less wax than I did and they had better alone is demonstrated by tools manufactured on 9 results. December 2003. Adhesive was placed on a stirring- This adhesive is distinctive under the micro- stick and heated directly over the fire; next, the scope; at 500! magnification, it is possible to see warm, pliable adhesive was used to attach the tools the fatty globules of wax in an otherwise homog- to their hafts, and the compound tools were heated enous mixture (Fig. 2c). The wax leaves a fatty directly over the fire for a few minutes. The stain on the stone insert, and this is visible even to mounting process was complete by 13:00 on the the naked eye. same day and the tools were thereafter allowed to The branches that were scraped with these tools air-dry naturally. Twenty-four hours later Haft 21 became reddened from the greasy coloring mate- (with adhesive made from 2.5 ml resin and 2.5 ml rial that rubbed against them. red ochre; see Table 1) was used to scrape bark. 4. Plant resin mixed with yellow coloring. Two The stone tool dislodged from its haft immediate- adhesives were manufactured using a yellow col- ly, revealing wet adhesive. The stone tool was oring material together with resin. When the pressed back into its haft and the exercise was adhesive was heated, it turned dark brown, not repeated on 12 December. The stone broke from red. No elemental analysis was attempted on the its haft after 1.5 min, once again revealing tacky yellow coloring, but it is possible that it is not an adhesive. The other tools from this experiment iron-rich mineral. Several coloring materials from were allowed to dry for a few more days before the Sibudu area were analyzed by S. Schiegl (by they were used. This exercise demonstrates that FT-IR; personal communication), who found that air-drying requires longer than four days, and some of the local yellow coloring material con- probably as long as six days, for the tools to tained no iron and yielded no infrared spectra become set in their adhesive, and for the adhesive information. to dry and harden properly. Neither of the tools hafted with adhesive made Tools mounted in adhesive on 6 January 2004 from yellow coloring and resin was successful. were processed differently after the first stage of heating the adhesive on a stirring-stick and The role of heat in the production of adhesive mounting the stone on the haft. They were slowly dried near a fire for between three and four hours. In the initial stages of adhesive manufacture, They were then used immediately to remove bark heating the adhesive helps to drive off excess and chop wood. All the tools were successful and moisture and make the mixture pliable. The use of none broke. L. Wadley / Journal of Human Evolution 49 (2005) 587e601 597

When using fire to dry adhesive, the tools’ Ochre-loaded adhesive is far easier to work with distance from the fire must be judged carefully. than sticky resin alone; it dries faster than When tools are too close to the fire the adhesive unloaded resin and, intuitively, it looks stronger, begins to boil. This creates cracks and weakness more homogeneous, and less brittle. Loaded resin and eventually the overheated adhesive produces can be more easily molded to a tool and it is less a huge bubble of air under the dry surface adhesive likely to fail during the drying process than natural crust. The adhesive can be rescued if it is quickly resin. In addition, unloaded adhesives are hydro- remolded and the tool placed next to a cooler part scopic and become tacky under damp conditions; of the fire. Slow drying in the vicinity of the fire ochre-loaded adhesive is not hydroscopic after it produces distinctive microscopic ‘‘mud cracking’’ has been properly dried. on adhesive loaded with red ochre. The strengths of my adhesives have not been empirically measured because the available meters Residue distribution on replicated tools are designed to measure the strength of metals without organic inclusions. At this stage, I also do All the replicated tools were hafted in such not have chemical or molecular results to explain a way that their proximal (thick, basal) ends were why the combination of iron-rich inorganic ochre secured to the hafts with adhesive. In most cases, with organic, carbohydrate-rich resin is an effec- the medial portions of the tools also contained tive bonding agent, but the tasks successfully adhesive. During the manufacture of the adhesive completed by tools mounted with ochre-loaded and mounting of the tools, my fingers became resin are proof of this. One clue to the success of stained with ochre and everything that I touched the ochre as a loading agent may be in nature: iron was marked with superficial red blotches. Thus, the oxide minerals act as cementing agents in some distal (tip) portions of the tools outside of the sedimentary rocks, such as banded ironstone; the hafted areas obtained ochre stains during the tool iron oxide is chemically deposited from solutions manufacturing process (see Fig. 1c). When the containing the mineral. Since (SiO2), like tools were used for chopping the bark off of iron oxide, is a cementing mineral in sedimentary branches, the superficial ochre stains on the distal rocks, it (or quartz-rich sand) should also prove to portions of the tools disappeared. They were be a good loading agent for adhesives. Empirical rubbed off through use, though the occasional research into the properties of ochre will be done spot of ochre remained. by T. Hodgskiss, who will also experiment with If the tool was placed on an L-shaped haft other loading agents because ethnographic and before adhesive was applied, the tool did not archaeological data imply that alternative loading acquire ochre-staining on the surface placed agents can be and have been used. Williamson against the haft. This observation is important (1997, 2004) occasionally observed fat, charcoal, because it means that archaeologically recovered and even blood together with resin on the bases tools with ochre on only one tool face should not (proximal ends) of both LSA stone tools from be discarded as unhafted. Other evidence for Rose Cottage and MSA stone tools from Sibudu hafting does, however, need to be sought. Cave, and plant material was a frequent inclusion in both instances. The Steenbokfontein Cave LSA adhesive used for mounting a contained Discussion plant fiber, but no ochre (Jerardino, 2001). My replications imply that artisans in the MSA My preliminary replication work on adhesives had considerable technical skill and that they and hafting confirms some of the observations understood the properties of the ingredients that made by other researchers. The work corrobo- are suitable for adhesive manufacture. They prob- rates, for example, the proposal by French ably manipulated ingredients knowingly, according researchers that ochre is excellent filler for use to their needs. Their complex technical knowledge with resins and with resin and wax mixtures. in all probability included an understanding of 598 L. Wadley / Journal of Human Evolution 49 (2005) 587e601 the individual properties of ochre and resin and Kranz Cave LSA is a good example of the combined property of these inorganic and the use of plant fiber for binding a tool to its haft organic materials. The adhesive makers would (Binneman, 1994). Lombard’s (2004, 2005) mi- have needed a good intuitive understanding of croscopy showed that binding was almost certainly how the adhesive should look and feel: experience also used on the Sibudu late MSA points because is required for successful adhesive manufacture it caused characteristic damage to the tool laterals. because the recipes are not an exact science. Plant residues at the base of tools might also Texture and viscosity must be gauged while the indicate that binding was used, but it is feasible adhesive is being mixed because the moisture that fragments of plant were intrinsic components content of resin varies from tree to tree and season of the glue that once attached the stone tool to its to season. The artisans also needed to understand handle or shaft. The replication work described the effects of various types of heating on the here suggests that plant residues can be uninten- adhesive recipes that they used. Furthermore, the tionally incorporated into resins during their creation of the hafting pastes needed planning to collection and processing. ensure that all the necessary ingredients were A comparison of the position of residues on the simultaneously available. This behavior is un- ancient tools from Sibudu Cave and on the doubtedly technically advanced. replicated hafted tools is informative. As men- In the same way that it is possible to tan hide in tioned above, the majority of the scrapers and different ways, it is also possible to haft tools in points examined had high frequencies of ochre different ways. Thus, several different types of residues on their proximal and medial portions, adhesive recipe could have been used for mounting although, in addition, small amounts of ochre are tools into hafts. To some extent, this variability sometimes present elsewhere on the tools. Almost may have depended on the local or seasonal half of the flakes examined also had ochre residues availability of ingredients. However, there is on their proximal and medial portions. It must be a strong possibility that some adhesive recipes pointed out that the ochre residues on MSA tools were deliberately varied to accommodate the type are often too small to be seen with the naked eye of task to be performed and the type of haft to be and are detected under a hand lens or binocular used. Microscopy on Rose Cottage and Sibudu microscope. The replicated hafted tools and the tools suggests that, on some occasions, the early archaeologically recovered tools that are assumed artisans loaded their resins, but under different to have been hafted have identical spatial distribu- circumstances, they used resin alone. As I have tions of ochre residues. The replicated tools do, shown, resin can be used alone, even though it is however, have heavier ochre stains around their difficult to manipulate when it does not have the proximal and medial portions, so that the staining addition of a loading agent. Unloaded resin has can be seen with the naked eye. The better visibility advantages as well as disadvantages. When resins is presumably because the replicated tools have not are heated alone, they are brittle and do not resist been buried; future work could involve experimen- high impact pressure (Rots, 2002: 57e59); they tal burying of tools with ochre residues. therefore allow barbs to become dislodged Archaeologists who believe that the mere in animal carcasses (Crombe´et al., 2001: 260). presence of ochre signifies symbolic behavior will This effect is desirable when projectile weaponry is probably contend that the use of ochre in adhesive used because it causes fatal hemorrhages. It is not is another example of ritual in the MSA. I cannot desirable when hand-held stabbing tools are used; prove that this was not the case; they may be right. here the requirement is different and points or For example, Hovers et al. (2003) argued persua- other spearhead inserts need to remain firmly in sively for the early symbolic use of ochre by their hafts so that the hunter can, when necessary, modern humans at Qafzeh. I do not yet have make repeated thrusts at prey. Securely fastened answers to all the questions about adhesive use spearheads may have depended on bindings as well and hafting practice. Nonetheless, there are data as loaded adhesives. The well-preserved Adam’s from several continents that verify that ochre has L. Wadley / Journal of Human Evolution 49 (2005) 587e601 599 properties useful for the manufacture of adhesive pieces might, on the one hand, have been smashed for hafting stone tools. This claim does not detract open to extract good quality coloring material and from ochre’s capacity as a symbolic agent, yet I there may be damage marks on the rocks that urge archaeologists to make interpretations that would confirm this practice. If, on the other hand, are sensitive to ochre’s versatility. Since ochre does large quantities of unused iron-rich rocks still have practical merit under certain circumstances, it retain good quality coloring material, we may be seems unwise to assume ritual use from its mere witnessing behavioral patterns that are not easily presence in archaeological sites. explained outside of a symbolic model. Ochre At the beginning of this paper, I commented research is by no means exhausted. that it is difficult to imagine what people living in At Rose Cottage and Sibudu Caves, it is perhaps the MSA would have done with the large unremarkable to find advanced technical knowl- quantities of coloring material that have been edge, and possibly even symbolic behavior, in found in some sites. The study reported here Howiesons Poort and post-Howiesons Poort in- suggests that hafting adhesives may account for dustries within the MSA. Such a comment is, a large proportion of the coloring. Some simple however, inappropriate when considering Lupem- calculations demonstrate the point. The three ban sites in Zambia, where quantities of coloring kilograms of coloring nodules used for my materials appeared at 300 ka together with the first replications produced sufficient hematite powder evidence for the presence of backed tools. Residue (about 70 ml) to make adhesives for 28 flakes. analysis results are not yet available for these Thus 107 g of geological parent rock were required Lupemban backed tools, which Barham (2002) to produce enough ochre powder for adhesive for believes may be the first examples of hafted tools. It one tool and 60 kg of parent rock would provide would be truly remarkable if hafting with ochre is enough powder for adhesive to haft 560 tools. discovered on these 300 kyr-old Twin Rivers backed Williamson’s (2004) study of more than 400 tools. While the presence of ochre-loaded tools may Sibudu stone tools (including flakes) showed that arguably represent symbolic behavior at 300 ka, it 11% had ochre on them. Even if only 5% of stone would certainly represent considerable and un- tools (including flakes) were hafted using ochre- expected technical skill by Homo rhodesiensis. rich adhesive, then 1000 tools from a collection of 20,000 might be hafted. This would require 107 kg of parent rock of the type that I used for my Acknowledgements experiments. The issue suggests that there is potential for further research into the amounts of I thank the NRF (National Research Founda- usable coloring material from a variety of iron-rich tion) for funding for the Sibudu project. The NRF geological rocks. The issue also raises the need for does not necessarily subscribe to the viewpoints standardization in the reporting of coloring expressed in this paper. I also thank the School of material from archaeological sites. Weights seem Geography, Archeology and Environmental Stud- necessary, as well as fragment frequencies, because ies for its support. I thank Paola Villa, Erella it is hard to compare 60 kg of material with 10,000 Hovers, and Marlize Lombard for their comments fragments. Another matter that seems important on a draft of this paper. Three anonymous to research is the reason for the appearance of reviewers also provided insightful comments that large quantities of unmodified coloring at MSA have helped to improve the paper; I am grateful sites. Watts (1998), for example, analyzed 4053 for their input. pieces of ochre from sites south of the Limpopo and found that only 384 pieces of ochre were definitely ground. We need an investigation of the References unmodified pieces from MSA sites. Are they unmodified because they are unusable or because Allain, J., Rigaud, A., 1986. De´cor et fonction: quelques exemles they were simply not used? Some unmodified tire´s du Magdale´nien. L’Anthropologie 90, 713e738. 600 L. Wadley / Journal of Human Evolution 49 (2005) 587e601

Allott, L., 2004. Changing environments in oxygen isotope Valladas, H., Watts, I., Wintle, A.G., 2002. Emergence of stage 3: reconstructions using archeological charcoal from modern human behavior: Middle Stone Age engravings Sibudu Cave. S. Afr. J. Sci. 100, 179e184. from South Africa. Science 295, 1278e1280. Ambrose, S.H., 1998. Chronology of the Later Stone Age and Henshilwood, C.S., Marean, C.W., 2003. The origin of modern food production in East Africa. J. Archaeol. Sci. 25, 377e392. human behavior: critique of the models and their test Audouin, F., Plisson, H., 1982. Les et leurs te´moins au implications. Curr. Anthropol. 44, 627e651. Pale´olithique en France: Enque´te et expe´riences sur leur Henshilwood, C.S., d’Errico, F., Vanhaeren, M., van validite´arche´ologique. Cahiers du Centre de Recherches Niekerk, K., Jacobs, Z., 2004. Middle Stone Age shell Pre´historiques 8, 33e80. beads from South Africa. Science 304, 404. Barham, L., 2002. Backed tools in middle central Hovers, E., Ilani, S., Bar-Yosef, O., Vandermeersch, B., 2003. Africa and their evolutionary significance. J. Hum. Evol. 43, An early case of color symbolism: ochre use by modern 585e603. humans in . Curr. Anthropol. 44, Beaumont, P.B., 1973. The ancient mines of southern 491e522. Africa. S. Afr. J. Sci. 69, 140e146. Jerardino, A., 2001. Diversity in mastic-mounted stone Beyries, S., Inizan, M.-L., 1982. Typologie, ochre, function. In: and the use of mastic in pre-colonial South Africa: evidence Cahen, D. (Ed.), Tailler! Pour quoi faire. Muse´royal de l’ from Steenbokfontein Cave. Antiquity 75, 859e866. Afrique centrale, Tervuren, pp. 313e322. Kimura, B., Brandt, S.A., Hardy, B.L., Hauswirth, W.W., Binneman, J., 1983. Microscopic examination of a hafted tool. 2001. Analysis of DNA from ethnoarcheological stone S. Afr. Archaeol. Bull. 38, 93e95. scrapers. J. Archaeol. Sci. 28, 45e53. Binneman, J., 1994. A unique stone tipped arrowhead from Knight, C., Powers, C., Watts, I., 1995. The human symbolic Adam’s Krans Cave, eastern Cape. South. Afr. Field revolution: a Darwinian account. Cambridge Archaeol. J. Archaeol. 3, 58e59. 5, 75e114. Binneman, J., Deacon, J., 1986. Experimental determination of Lombard, M., 2004. Distribution patterns of organic residues use wear on stone adzes from Boomplaas Cave, South on Middle Stone Age points from Sibudu Cave, KwaZulu- Africa. J. Archaeol. Sci. 13, 219e228. Natal, South Africa. S. Afr. Archaeol. Bull. 59, 37e44. Bleek, W.H.I., Lloyd, L.C., 1911. Specimens of Bushman Lombard, M., 2005. Evidence of hunting and hafting during the Folklore. George Allen and Company Ltd., London. Middle Stone Age at Sibudu Cave, KwaZulu-Natal, South Bu¨ller, J., 1988. Handling, hafting and ochre stains. In: Beyries, S. Africa: a multianalytical approach. J. Hum. Evol. 48, (Ed.), Industries Lithiques Trace´ologie et Technologie Vol. 1: 279e300. Aspects arche´ologiques. BAR International Series, 411 (i), Mandl, I., 1961. Collagenasis and elastases. Adv. Enzymol. 23, Oxford, pp. 5e32. 164e264. Clark, J.D., Brown, K.S., 2001. The Twin Rivers Kopje, Marshall, L., 1976. The !Kung of Nyae Nyae. Harvard Zambia: stratigraphy, fauna, and assemblages from University Press, Cambridge. the 1954 and 1956 excavations. J. Archaeol. Sci. 28, 305e330. McBrearty, S., 2001. The Middle Pleistocene of East Africa. In: Crombe´, P., Perdaen, Y., Sergant, J., Caspar, J.-P., 2001. Wear Barham, L., Robson-Brown, K. (Eds.), Human Roots: analysis on early from the Verrebroek Africa and Asia in the Middle Pleistocene. Western Site, East Flanders, . J. Field Archaeol. 28, 253e269. Academic and Specialist Press Ltd., Bristol, pp. 81e98. Deacon, H.J., 1979. Excavations at Boomplaas Cave: a se- McBrearty, S., Brooks, A.S., 2000. The revolution that wasn’t: quence through the Upper Pleistocene and in a new interpretation of the origin of modern human South Africa. World Archaeol. 10, 241e257. behavior. J. Hum. Evol. 39, 453e563. Deacon, H.J., 1995. Two late PleistoceneeHolocene archeo- Mitchell, P., 1995. Revisiting the Robberg: new results and logical depositories from the southern Cape, South Africa. a revision of old ideas at Sehonghong , S. Afr. Archaeol. Bull. 50, 121e131. . S. Afr. Archaeol. Bull. 50, 28e38. Deacon, H.J., Deacon, J., 1980. The hafting, function and Phillipson, D.W., 1976. The of Eastern Zambia. distribution of small convex scrapers with an example from British Institute in Eastern Africa, Nairobi. Boomplaas Cave. S. Afr. Archaeol. Bull. 35, 31e37. Pomie` s, M.-P., Menu, M., Vignaud, C., 1999. Red Palaeolithic Flood, J., 1995. Archeology of the Dreamtime: The Story of pigments: natural hematite or heated goethite? Archaeome- Prehistoric Australia and Its People. Angus and Robertson, try 41, 275e285. Sydney. Rots, V., 2002. Hafting traces on flint tools: possibilities and Gibson, N.E., Wadley, L., Williamson, B.S., 2004. Residue limitations of macro- and microscopic approaches. Ph.D. analysis of backed tools from the 60 000 to 68 000 year-old Dissertation, Katholieke Universiteit Leuven. Howiesons Poort layers of Rose Cottage Cave, South Schweitzer, F., 1979. Excavations at Die Kelders, Cape Africa. South. Afr. Humanit. 16, 1e11. Province, South Africa: the Holocene deposits. Ann. S. Grant, R., Thomas, V., 1998. Sappi Tree Spotting: KwaZulu- Afr. Museum 78, 101e232. Natal. Jacana, Johannesburg. Thackeray, A.I., 2000. Middle Stone Age artifacts from the Henshilwood, C.S., d’E´rrico, F., Yates, R., Jacobs, Z., 1993 and 1995 excavations of Die Kelders Cave 1, South Tribolo, C., Duller, G.A.T., Mercier, N., Sealy, J.C., Africa. J. Hum. Evol. 38, 147e168. L. Wadley / Journal of Human Evolution 49 (2005) 587e601 601

Tomlinson, N.E., 2001. Residue analysis of segments, backed Wadley, L. A typological study of the final Middle Stone Age and obliquely backed blades from the Howiesons Poort stone tools from Sibudu Cave, KwaZulu-Natal. S. Afr. layers of Rose Cottage Cave, South Africa. M.A. Thesis, Archaeol. Bull., in press. University of the Witwatersrand. Wadley, L., Jacobs, Z., 2004. Sibudu Cave, KwaZulu-Natal: Van Wyk, B.-E., Gericke, N., 2000. People’s Plants: A Guide to background to the excavations of the Middle Stone Age Useful Plants of Southern Africa. Briza Publications, and Iron Age occupations. S. Afr. J. Sci. 100, Pretoria. 145e150. Velo, J., 1984. Ochre as medicine: a suggestion for the Wadley, L., Lombard, M., Williamson, B., 2004. Ochre as interpretation of the archeological record. Curr. Anthropol. a hafting material of Middle Stone Age tools: the evidence 25, 674. from Sibudu Cave, KwaZulu-Natal, South Africa. Antiq- Villa, P., Delagnes, A., Wadley, L., 2005. A late Middle Stone uity 78, 661e675. Age artefact assemblage from Sibudu (KwaZulu-Natal): Watts, I., 1998. The origin of symbolic culture: the Middle comparisons with the European Middle Palaeolithic. Stone Age of southern Africa and Khoisan ethnography. J. Archaeol. Sci. 32, 399e422. Ph.D. Dissertation, University of London. Wadley, L., 1987. Later stone Age Hunters and Gatherers of Watts, I., 2002. Ochre in the Middle Stone Age of southern the Southern Transvaal: Social and Ecological Interpreta- Africa: ritualized display or hide preservative? S. Afr. tions. BAR International Series, Oxford. Archaeol. Bull. 57, 1e14. Wadley, L., 1997. Where have all the dead men gone? Stone Webley, L., 1994. A short note on the Namaqualand diary Age burial practices in South Africa. In: Wadley, L. (Ed.), entries of W.G. Atherstone relating to Bushman bows and Our Gendered Past: Archeological Studies of Gender in arrows. Southern Afr. Field Archaeol. 3, 60e62. Southern Africa. Witwatersrand University Press, Johan- Williamson, B.S., 1997. Down the microscope and beyond: nesburg, pp. 107e134. microscopy and molecular studies of stone tool residues and Wadley, L., 2001. What is cultural modernity? A general view bone samples from Rose Cottage Cave. S. Afr. J. Sci. 93, and a South African perspective from Rose Cottage Cave. 458e464. Cambridge Archaeol. J. 11, 201e221. Williamson, B., 2004. Middle Stone Age tool function from Wadley, L., 2003. How some archaeologists recognise culturally residue analysis at Sibudu Cave. S. Afr. J. Sci. 100, 174e modern behaviour. S. Afr. J. Sci. 99, 247e250. 178. Wadley, L., 2004. Seeds from the Middle Stone Age of Sibudu Wurz, S., 1999. The Howiesons Poort at Klasies Riverdan Cave: the environmental implications. S. Afr. J. Sci. 100, argument for symbolic behaviour. S. Afr. Archaeol. Bull. 167e173. 54, 38e50.